- How do you know that there are more conflict-related deaths than have been reported to the Commission for Reception, Truth and Reconciliation (CAVR, by its Portuguese acronym)?
- Where did the method of multiple systems estimation come from?
- If you didn’t have access to the whole population, how do you know how representative these data are of the entire population? i.e. How do you control for bias?
- What are the total conflict-related mortality numbers? How many people were killed and disappeared between 1974 and 1999? And how many people died due to hunger and illness?
- What is the margin of error associated with these results?
- What is unique about these estimates?
- Are the combatants (e.g. Indonesian soldiers, officers, and police officials, as well as the resistance fighters) who died in the conflict included in the estimate of 18,600 killings?
- How could so many deaths have been unknown?
- What is the upper limit estimate of conflict-related deaths? What does it mean?
- What was the extent and pattern of conflict-related displacement in Timor-Leste between 1974 and 1999?
- Who was responsible for the large-scale, conflict-related displacement in Timor-Leste?
- What was the relationship between conflict-related mortality and conflict-related displacement?
- Why collect the human rights violation data and develop this sort of analysis, what is the purpose of this work?
- What are the future prospects for this work?
Conflict-related mortality in Timor-Leste (East Timor) consists of two parts:
(i) famine-related deaths (i.e. deaths due to illness and hunger in excess of the total that would be expected if the death rate due to hunger and illness had continued as it was in the pre-invasion peacetime period.), and
(ii) political violence deaths (i.e. killings and disappearances).
The CAVR did not document all famine-related deaths and political violence deaths, only some of them. The CAVR’s statement-taking process reflects the experience of 7,688 respondents, but approximately 940,000 other East Timorese did not give their testimonies. The CAVR’s Retrospective Mortality Survey (RMS) reflects the experiences reported in 1,396 households, but omits the experiences of nearly 190,000 households not sampled. Furthermore, the CAVR was not able to enumerate every gravestone in Timor-Leste (for example, Chinese graveyards, private burial plots and some public graveyards were omitted).
However, even if the CAVR’s statement-taking process and RMS did reflect the experience of every living person in Timor, many deaths would remain undocumented because all people who could remember those particulars may have died, left the country or were unable to recount the stories during CAVR’s data collection period. Social memory is always partial. Our estimates are based on the total number of deaths which could be remembered by survivors resident in Timor-Leste in 2004. Both our estimation methods, using the RMS and Multiple Systems Estimation, are substantially conservative because many deaths could not be remembered by 2004.
Multiple-systems estimation is commonly used to correct censuses and to estimate the size of wildlife populations. Benetech’s® Human Rights Data Analysis Group has used this method to estimate total deaths due to political violence in Guatemala (on behalf of the Commission for Historical Clarification), in Kosovo (as part of expert testimony presented in the trial of Slobodan Milošević) and in Perú (on behalf of the Commission for Truth and Reconciliation).
When we take testimonies, we don’t know if the statistics created from the testimonies represent the reality of the social world. For example, if we interview too many people in the urban areas and not enough in rural areas, we end up underestimating the percentage of the deaths that happened in the rural areas. More dangerously, we may end up with biased statistics if people with a certain set of characteristics perceive the CAVR as an instrument of the state, and are therefore less likely to report violations attributed to the Timorese resistance movement. The estimates we created are designed to correct biases that might result from unequal coverage. Specifically, we designed the modeling and stratification to control for biases that would affect our estimates that compare different regions. By estimating the number of unknown deaths, we can analyze the total magnitude of conflict-related deaths, for different regions and for different phases of the conflict. This is the only way to know how many Timorese died as a result of the conflict.
We estimate the lowest possible number of conflict-related deaths during the CAVR’s reference period, 1974-1999, is 102,800. Of these estimated 102,800 deaths, approximately 18,600 Timorese were killed or disappeared, while approximately 84,200 died due to hunger and illness in excess of what would be expected due to peacetime mortality.
Because these are estimates, there is some uncertainty associated with them. This uncertainty is expressed by a margin of error.
Each of the estimates has an individual 95% confidence interval shown in the results (in the graphs and tables). The overall estimate (102,800) has a margin of error of +/- 12,000. The estimated 18,600 killings have a margin of error of +/- 1,000. The estimated 84,200 deaths due to hunger and illness which exceed the total that would be expected if the death rate due to hunger and illness had continued as it was in the pre-invasion peacetime period, have a margin of error of +/- 12,000.
These estimates identify both the total magnitude of conflict-related mortality and the proportional share of cause-specific mortality. The estimates, themselves, are based on three new, independent datasets and standard demographic and statistical methods. Previous attempts to estimate conflict-related mortality were based on intuitive estimates by informed observers and indirect estimates based on population censuses. Most previous estimates did not distinguish between political violence deaths and famine-related deaths.
No. Although some combatant deaths were reported in the CAVR’s statement-taking process and Retrospective Mortality Survey, these data sources did not have sufficient coverage of combatant deaths to produce valid statistical estimates of combatant deaths. Given the focus of CAVR’s mandate on human rights violations and the Commission’s limited time and resources, the CAVR decided not to include killings of combatants in its mortality estimates.
Widening the scope of statistical estimates to include “combatant” deaths would have required large-scale data collection work in West Timor and other parts of Indonesia where Indonesian military and pro-autonomy militias were located in 2002-2004 when the CAVR was collecting data. The CAVR initially did try to collect data in West Timorese refugee camps, and managed to collect 86 interviews within the camps. However, due to security concerns for CAVR staff and interviewees, the Commission was forced to halt these data collection efforts.
Since we did not obtain high levels of interviewee reporting about killings of combatants during the CAVR narrative testimony collection process (referred to in our report as the Commission’s Human Rights Violations Database, “HRVD”) or the Retrospective Mortality Survey, we were unable to make estimates for killings of combatants.
Most of the conflict-related deaths in Timor-Leste occurred in the late 1970’s and early 1980’s when access to Timor-Leste was very limited. At that time, human rights groups, humanitarian agencies and the international community were only able to document a portion of the political violence and famine-related deaths which occurred.
We estimated that the total deaths due to hunger and illness, in excess of a Crude Death Rate (CDR) baseline, could be as high as 183,000. This upper-bound estimate adjusts for possible underestimation resulting from the loss of social knowledge. That is, the upper-bound estimate takes into consideration the magnitude of hunger and illness deaths which left no surviving parents, siblings or children who could have been respondents in the CAVR’s Retrospective Mortality Survey.
The upper-bound estimate of 183,000 deaths due to hunger and illness are based on a number of assumptions, including assumptions about the shape of the decline of the Crude Death Rate (CDR) from the early 1970s through the late 1990s and about the nature of the loss of social memory. This estimate is also subject to substantial sampling and non-sampling error.
The unadjusted estimate of 102,800 conflict-related deaths must be a minimum bound, as it only represents those deaths which could be remembered by people resident in Timor-Leste during the CAVR’s data collection. This is a subset of the total number of conflict-related deaths which actually happened. While the upper-bound estimate does control for loss of social memory, increased uncertainty and error is associated with this estimate. We recommend that the appropriate and conservative finding is that there was a minimum of 102,800 deaths in excess of the peacetime baseline.
Conflict-related displacement was widespread. We estimated that 108,200 (+/- 7,187) households experienced 282,800 (+/- 31,100) displacement events between 1974 and 1999.
Most displacements occurred between 1975 and 1980. The maximum years are 1975 and 1976, with 61,400 (+/- 13,300) and 59,800 (+/- 7,200) displacement events, respectively. The events of 1999 were substantially fewer, with approximately 28,100 (+/- 5,600) events.
Most displacements were local. Of all displacement events, 54.3% are within subdistrict, 15.6% are within district, 17.4% are within region, 9.3%% are within East Timor, and 2.4% are outside of Timor. However, in 1999, the displacements that take the household out of East Timor increase to 19.3% (+/-6.1%) of displacements in that period.
The institution that respondents reported most frequently as the group telling them to move was the Indonesian military (46.4%), followed by Fretilin’s military wing, Falintil (15.0%) and Timorese militias backed by the Indonesian military (8.8%). Respondents reported that ‘conflict’ motivated 52.3% of their displacements, with ‘forced by Indonesian military’ contributing an additional 16.3%.
The pattern of conflict-related displacement and conflict-related mortality (i.e. both famine-related deaths and political violence deaths) are positively correlated over time and space. On average, when conflict-related displacement increases (or decreases) at a particular time or in a particular region, conflict-related mortality also increases (or decreases). These phenomena are therefore likely to have a common cause. The pattern of rapid increase in killings and disappearances, deaths due to hunger and illness and displacement at the beginning of the Indonesian occupation is consistent with the claim that the occupation caused the increased mortality in Timor-Leste.
The magnitude of human rights violations in Timor-Leste during the Indonesian occupation and responsibility for these violations has long been a subject of contentious debate. The Indonesian military has claimed that they were not responsible for the majority of violations during their occupation of Timor-Leste, human rights advocacy groups have argued otherwise. By establishing an empirical basis for this debate, CAVR and HRDAG have helped to shed new light on these important questions about policy, practice and responsibility. On January 20th, 2006, Juwono Sudarsono,the Indonesian Defense Minister, responded to these findings by saying, “This is a war of numbers and data about things that never happened…”. We hope that our findings and the publication of the anonymized data will promote a debate about accountability for the past amongst the Timorese people, Indonesian government and international community. In particular, we invite the Indonesian government’s statisticians to review our work and engage in an open, technical debate about our findings and methods.
We have publicly released our statistical report to the CAVR which details widespread and systematic violations in Timor-Leste. In accordance with Benetech’s Memorandum of Understanding with the CAVR, we have also published these data on the Internet so that human rights researchers, statisticians and demographers can use the material to replicate our findings and continue research on past human rights violations in Timor-Leste. We plan to release extensions to our analysis over the next few years, to ensure that the ongoing debates about truth and accountability for past human rights violations are informed by these important data sources.